Addicted to Talk: Newspaper Representations of the Female Speaking Subject

Transcription

1 1. Introduction Addicted to Talk: Newspaper Representations of the Female Speaking Subject Tony Fisher University of Nottingham The intersection of gender and the representational practices of the mass media has long been a preoccupation not only for media researchers (MacDonald 1995, Gauntlet 2002, Gill 2007), but also a number of linguists associated with language and gender research: scholars such as Caldas-Coulthard (1996), Cameron (2003, 2007) Litosseliti (2006), and Sunderland (2004, 2006, 2010; Sunderland & Taylor 2003), for whom the media are seen as a key site for the discursive construction of gendered identities and maintenance of dominant gender ideologies. The present study examines specifically how the relationship between language and gender is constructed in media texts, and to this end interrogates a small sample of articles drawn from the online editions of the most widely read national newspapers of the United Kingdom. The study describes the range of language and gender discourses that can be observed in the dataset, while also examining the linguistic construction of one specific language and gender discourse in one particular text. Section 2 sets the scene with a consideration of newspapers as economic, sociocultural and discourse practice, with a particular emphasis on the changing modes of consumption heralded by the advent of the internet. This is followed in section 3 by an account of the theoretical ground upon which the paper is based, before turning in section 4 to a brief explication of the rationale for focusing on transitivity. Section 5 sets out the methods employed in both data collection and analysis, while the analysis itself is presented in section 6. The analysis is presented in two sub-sections, the first a content based analysis of the entire dataset, the second an analysis of transitivity relations in one text from the dataset. This is followed in section 7 by a consideration of the wider social significance of the findings reported in the preceding analysis for women as media consumers, and concludes in section 8 with some final reflections on the value of linguistic analysis in challenging dominant language and gender discourses and the broader gender ideologies they articulate. It is important to state at the outset that, as is the case with most language and gender research, the present study is conducted from a committed feminist perspective, and consequently draws on a range of feminist scholarship, mostly though not exclusively associated with linguistics and its various sub-disciplines and approaches. Some readers may object to this position on the grounds that an author conducting politically motivated research does so with a priori assumptions, informed by the theoretical positions with or against which they align themselves, a practice that has elicited a charge of theoretical imperialism from Emanuel Schegloff (1997:167). It is hoped that the mixed methods approach employed in the present study will, to some extent at least, serve as a corrective in this regard, in that the quantitative analysis provides an empirically grounded warrant for the focus of qualitative analysis that follows. However, it must be stressed that this study is not intended as an Tony Fisher, Addicted to Talk: Newspaper Representations of the Female Speaking Subject, Working With English: Medieval and Modern Language, Literature and Drama 7 (2011), pp

2 Newspaper Representations of the Female Speaking Subject 15 exercise in simply describing language and gender discourses, but is offered as a praxisoriented critical analysis (Lazaar 2005: 6) that aims to make a small contribution towards fulfilling the political goal of aiming to redress gender imbalances and move a step closer towards bringing about gender emancipation and equality (Mills & Mullany 2011: 4). 2. Newspapers and the online revolution Newspapers, both regional and national, are a well-established daily source of textual interaction for many people. Roger Fowler notes the habituality of consumption that results from the daily modes of production and distribution of newspapers, arguing that for most people reading the daily newspaper makes up their most substantial and significant consumption of printed discourse. For the majority it is second only to television as a window on the world (1991: 121). Fowler goes on to argue that newspapers are thus imbued with a major ideological importance that derives much of its character from the commercial practices within which newspaper production is constituted. The text from which these comments are taken was published in 1991, somewhat before the advent of the internet, and the intervening years have witnessed nothing short of a communications revolution. By the time the present study was conducted it was the case that every national UK newspaper maintained an online edition, and that while the popularity of online newspapers seemed to be growing, sales of print editions was clearly in decline (Waldman 2007). Researchers interested in new media have noted that the majority of the news stories and feature articles found online are identical to those that appear in print editions (Kenney, et al. 2000), but there are nonetheless a number of important differences between print and online content. While most material published online by newspapers also appears in print, it is entirely possible that print and online versions of articles may undergo different editorial transformations, resulting in subtle differences between online and print versions. A further consideration is the interactive nature of online news content, manifest in hyperlinks to related stories and to advertisers websites, and in the increasingly common practice of providing a readers feedback channel after selected articles. Online newspapers differ from print editions in another important way, in that while print editions are available (with some limited exceptions) only in the country of origin and only on the day of publication, online news content is globally accessible for as long as it is retained in the online archive maintained by the newspaper in question. This obviously makes online news a far more accessible data source for the analyst, but it also means that newspaper articles published online have the potential to remain ideologically active over a far wider area and for far longer than those that appear only in print. For these reasons, the choice of attention to online content over printed material is not simply a matter of accessibility, but an attempt to focus on relevant news content in its most potentially wide-reaching form, while capitalizing on the increased potential for exposure to critique, by analyst and interested reader alike, that is a necessary consequence of this mode of publication. 3. Language and Gender Ideologies and Discourses Mills notes that cultural and critical theory in the 1990s and 2000s was largely characterised by intense theoretical difficulty (2005: 26) in deciding whether to draw on discourse or ideology as the primary concept upon which to base analysis of the social world. It is tempting to think of the difficulty Mills describes as a moment in a general transition from Marxist-inflected views of the social to a more nuanced, post-marxist approach fuelled by growing interest in discursive practices in institutions across a range of disciplines, and influenced by post-war French discourse analysis, in particular the work of Michele Foucault. It is certainly true that Marxist theory has met with stiff competition from discourse theory

3 16 Tony Fisher during the last 4 decades. For feminist scholars in particular, the concept of discourse has proved politically expedient, since the Foucauldian conceptualisation of power with which it is normally associated allows for the possibility of resistance in a way that is difficult within the structural Marxism of Louis Althusser (1984), where ideology receives one of its earliest and most influential treatments. However, that ideology has been replaced as a term of art is clearly not the case, as it continues to be used widely alongside discourse in the work of scholars associated with a variety of fields and approaches. Having said that, it is sometimes possible to discern a tendency for individual researchers to privilege one concept over the other, a tendency that will be illustrated with regard to language and gender scholarship below. In a 2003 position paper, Cameron builds upon work carried out in the emerging subdiscipline of language ideologies, a field of inquiry that has its origins in North American Linguistic Anthropology and that seeks to account for representations, whether explicit or implicit, that construe the intersection of language and human beings in a social world (Woolard 1998: 3). In this paper, Cameron discusses what she calls the new deficit model, according to which women and girls are portrayed as superior language users, while men and boys are regarded as deficient in this regard. Cameron s deficit model is new because it runs counter to the original deficit model associated with second wave feminist scholars, most notably Robin Lakoff (1975), in which the roles were the reverse. However, Cameron argues that it is mistaken to interpret these new representations of women as a feminist gain, since the new deficit model is in fact an ideological construct that can be explained in terms of a more general realignment of linguistic values characterised by an increased emphasis within public discourse on the private speech genres of conversation and self-disclosure stereotypically and historically associated with (some) women (2003: 461). Thus, while the new deficit model constructs women as superior language users, in doing so it merely reinforces and naturalises the gender dichotomy on which it is founded, while simultaneously maintaining a gender hierarchy in problematizing a situation in which women are seen to be gaining an advantage at the expense of men. A set of representational tendencies similar to those identified by Cameron are also commented on by Sunderland (2004: 43), but whereas Cameron s focus is on gender and language ideologies, Sunderland sees such representations as manifestations of a poor boys discourse (emphasis mine), one of a number of gendered discourses she finds to have been identified in the research literature. On the face of it there seems to be little difference between Sunderland s use of the term discourses and Cameron s of ideologies, and indeed Sunderland notes that it is not uncommon to find the two terms used interchangeably. Both concepts are frequently defined in terms of representations see in particular Fairclough (2003), who in the same volume defines ideologies and discourses as representations of (2003: 9) and ways of representing aspects of the world respectively (2003: 124, emphasis mine). Sunderland, who like Fairclough writes from a Critical Discourse Analysis (CDA) perspective, goes as far as to argue that it is possible to view ideology as the cultural materialist antecedent of the post-structuralist use of discourse, but the fact that discourse and ideology are distinct for Sunderland is clear when she adds that for both poststructuralism and CDA, discourse can be seen as carrying ideology (2006: 6, emphasis in original). A similar view is evident in Fairclough (1992), who sees ideologies as being embedded in discursive practices (1992: 87), and van Dijk, who sees discursive structures as phenomena that can be used to express ideological belief (ibid: 157). However, while Sunderland s conceptualisation of the two phenomena in terms of carrier and carried suggests a view of ideologies as essentially prediscursive phenomena, van Dijk highlights the mutually constitutive nature of the discourse/ideology dichotomy when he speaks of ideologies not only as influencing text and talk, but also developed by or through discourse (1995: 135).

4 Newspaper Representations of the Female Speaking Subject 17 A different view again of the ideology/discourse relation is proposed by Eagleton, who notes that the traditional Marxist conceptualisation of ideology in terms related to (false) consciousness overlooks the fact that consciousness is itself a reification, an abstraction from our actual forms of discursive practice (1991: 193). While Althusser s structural Marxism offers some degree of corrective in this regard, Eagleton notes that Althusser risks shifting the emphasis too far in the opposite direction, thereby reducing concepts to social practices (ibid: 194). Eagleton s solution is to posit a third way in which ideology is viewed less as a particular set of discourses, than as a particular set of effects within discourses (ibid). But if ideologies are effects within discourses, then it follows that it is only by identifying and accounting for discourses that ideologies can be observed. In this sense, discourses can be viewed as the material discursive relations and structures in which ideologies inhere, and it is with precisely such an understanding of the relationship between the two that the terms discourse and ideology are employed in the present study. The relevance of language ideologies in particular to the present study is clearly evident, and indeed the growing concern with language ideologies in popular culture has resulted in the recent publication of two collected editions (Johnson & Ensslin 2007, Johnson & Milani 2010), both of which focus specifically on representations of language in the mass media. Like Cameron s analysis, however, the present study is concerned not only with representations of language, but also gender, and specifically with media representations of the relationship between the two. While Cameron is concerned primarily with ideologies, the present study follows Sunderland in taking as its immediate analytical focus discourses, but the concept of ideology is nevertheless retained. Indeed, as Mills notes, discourse has been defined in dialogue with and in relation to the definition of ideology (2005: 28), and in this sense one might go as far as to say that it is almost impossible to talk about either concept without reference to the other. Both discourse and ideology are problematic, but if one accepts that the two refer to different and distinct phenomena that are at least partially defined in relation to one another, then it follows that it is not only possible, but actually necessary to draw on both concepts in any critical analysis of language and text. As will hopefully be clear from the discussion above, the present study, while not necessarily wishing to declare complete allegiance to either language ideologies or CDA, draws on both, particularly for the definitions of discourse and ideology that underpin the analysis below. The analysis that follows draws on Eagleton s view of ideology as a set of effects within discourses, but places additional emphasis on ideologies as structures not only articulated in discourse, but also partially constituted in the discursive moment. Such a view necessarily requires a consideration of agency and the role of the reader (and analyst) in the (re)production of discourses and ideologies when consuming media texts, a point that will be discussed below. 3.1 Discourses and Statements The task of identifying language and gender discourses necessitates a consideration of precisely what it means to talk about a discourse and how it might be possible to identify those individual discourses invoked or drawn upon in any given text. The difficulties in defining discourse have been widely noted, and Foucault himself admits to treating it sometimes as the general domain of all statements, sometimes as an individualizable group of statements, and sometimes as a regulated practice that accounts for a certain number of statements (1972: 80). Common to all three of Foucault s formulations is the notion of the statement, so that, as Fairclough notes, the analysis of discourse for Foucault is the analysis of the domain of statements (2003: 123). The statement, or enoncé in the original, has been subject to a variety of definitions and interpretations, but a particularly useful starting point is that offered by Mills (1997), who takes the statement to be the primary building

5 18 Tony Fisher block of a discourse (1997: 54), describing statements as those utterances and texts which make some form of truth-claim [ ] and which are ratified as knowledge (1997: 55). Foucault himself explicitly rejects the conceptualisation of the statement as the atom of discourse, arguing that it is not in fact equivalent to a structural unity such the sign at all, but rather a function that cuts across a domain of structures and possible unities (1972: 98). This emphasis on the statement as a function of the sign also runs contrary to Mills association of the statement with the utterance or text, since an utterance may function as more than one statement, while a single statement may be realised in any number of discrete utterances or signs (Dreyfus & Rainbow 1983: 45). However, while not analogous to the sign, statements are certainly dependent upon signs for their material realisation, and it is by focusing on utterances in terms of the assertions they seem to make, and the grounds upon which they make them, that one can begin to identify the enunciative force of utterances and texts. Furthermore, statements exist not in isolation, but rather derive their enunciative functionality and character from the relations into which they enter with other statements, relations in which regularities can sometimes be observed (Foucault 1972: 38), and discourses thereby identified. Of course, the identification of statements and their organisation into discourses is an inherently interpretive process, a point emphasised by Sunderland in her characterisation of discourses as either descriptive or interpretive categories (2004: 6). The interpretive nature of ideology is noted by Mills who, drawing on a broadly post-structuralist model of text, argues that ideology cannot be viewed as a unitary entity, but rather as something that is negotiated by individual agents and which individual subjects will negotiate, affirm, and/or resist, (1995: 149), and also by Fairclough who notes that meanings are produced through interpretations of texts, and texts are open to diverse interpretations which may differ in their ideological import (1992: 89). It seems not unreasonable therefore, particularly considering the convergence in use of the terms discourse and ideology discussed above, to assume that the enunciative functions that particular statements come to possess, and therefore also the discourses into which they will be seen to coalesce, are the result of a process of negotiation during which readers draw not only on (inter)textual cues and traces, but also on their personal habitus, the system of structured and structuring dispositions (Bourdieu 1990: 52) that result from an individual s discursive and social histories, and that predispose the individual towards certain interpretations or readings of text and other social phenomena over others. This is an important point, since it means that the discourses identified and discussed in the analysis below, and the enunciative functions identified in the stretches of texts from which such discourses are seen to emerge, are inherently interpretive, a result as much of the expectations and predispositions of the author of this research paper as of anything empirically observable in the texts examined. It is with this caveat in mind that in section 5 I will account for the precise ways in which discourse theory was applied in both the collection and analysis of data in the present study. But before we move on to consider the practicalities of the theory-driven methodological approach taken, it will first be useful to consider briefly the rationale behind the decision to focus on transitivity in the main qualitative component of the analysis below. 4. Why transitivity Analysis? It is over thirty years since Halliday first developed the model of transitivity that is now associated with Systemic Functional Grammar, and it is fair to say that in that time Halliday s basic model has remained largely unchallenged and unchanged. So why focus on transitivity? What can attention to this system reveal that isn t already widely documented? In fact, transitivity is explored in the present study largely because this is what emerged from exploratory engagement with the data. However, where transitivity analysis frequently

6 Newspaper Representations of the Female Speaking Subject 19 focuses on the referential dimension of the ideational metafunction to which it belongs (Walsh 2001: 56), the present analysis focuses specifically on the evaluative dimension, revealing how the transitivity relations evident in the text serve to represent women in particular and ideologically significant ways. In this it will be demonstrated how attention to the evaluative aspect of the ideational metafunction, and transitivity in particular, can yield useful insights into the ideological significance of the choices made by text producers, choices that are often motivated by a desire to position listeners/readers as compliant subjects (Walsh 2001: 31). While not a central preoccupation in language and gender research, transitivity has attracted the attention of a number of scholars working in the field. Mills (1995) examines the significance of transitivity choices in the representation of gendered social actors in her analysis of literary texts. In one example, Mills highlights a significant discrepancy in the number of clauses in which the male and female character are agentive (14 and 5 respectively), while also pointing out that while the male character is agentive in all 10 of the material processes predicated of him, the female character is agentive in none. Mills analysis thus points to the potential utility of transitivity analysis in representations of women as passive rather than agentive, acted upon rather than acting upon others, and the ideological readings to which literary and non-literary texts may be open in this regard. In another study, Clark (1998) examines transitivity relations in articles published in The Sun newspaper. Clark analyses a series of reports of violent crimes against women, finding that, through a combination of naming (or more accurately labeling) practices and transitivity choices, the blame attributed to male attackers for the violent acts reported is either backgrounded, or effectively transferred to the female victims. In a particularly striking example, Clark demonstrates how the in the headline GIRL 7 MURDERED WHILE MUM DRANK AT THE PUB, not only is the murderer made invisible by deletion, but a causal relationship is established in the text between the mother s absence and the child s death, suggesting a reading in which the child died because her mother was out (1998: 188). What these two studies show is that the analysis of transitivity and gender is sometimes, but not always, a matter of highlighting the repeated representation of women as passive, and more or less powerless against the agentive actions of men. In particular, Clark demonstrates that women frequently are constructed as agentive actors, but her analysis suggests that the significance of women s agency may often differ fundamentally from that attributed to men. This is an observation that finds support in the analysis below. 5. Method The study interrogates a corpus of articles drawn from British national newspapers between 2006 and The texts that comprise the dataset were all found using the search engines embedded within the individual newspaper websites, employing a set of search terms designed to locate articles in which discourses relating to language and gender were likely to be evident. The search terms used comprised one nominal item referring to a gendered human category, and one verbal element relating to a verbal process. Both singular and plural forms of the nominals and all inflected forms of the verbs were used, giving rise to search terms such as women talk, women discussing, boy chat and men gossiped. Also included were search terms comprising the elements language/talk and gender/sex. This procedure resulted in a final dataset of 47 texts, comprising just over 64,000 words of text. An integrated mixed-methods approach is taken throughout, beginning with a content analysis of the entire dataset, in which the relevant discourses and their constitutive statements are both identified and described. Any stretches of text that were considered, either explicitly or implicitly, to make some kind of assertion about the relationship between language and gender, and that could therefore be considered to possess the potential to

7 20 Tony Fisher function as statements in the Foucauldian sense, were identified and catalogued. Only the first instantiation of any particular language and gender discourse in any single text was counted, so that the quantities indicated are the number of texts in which the various discourses identified were observed at least once. The textual elements thus identified were recorded along with an interpretive note that sought to identity their likely dominant enunciative function, a process that was repeated several times in order to ensure as far as possible the validity of the interpretations. Using these notes, points of similarity and contrast were identified, and the stretches of text arranged into groups that were interpreted as displaying similar or related enunciative functions. In this way, the elements of text identified were categorised into a series of discursively distinct groupings, emanating in part form the data, in part from those categories identified as significant within language and gender research as an academic discipline (Foucault 1981), and in part from the expectations and predispositions that are a function of my own personal habitus. Utterances that seemed to contradict one another were assigned to the same category if they were considered to be ideologically divergent or opposing representations of a common discursive object, as in the case of utterances that were seen to support and challenge the stereotype of the talkative women. The co-classification of such contradictory elements is a necessary consequence of the unevenness of discourses observed by Mills when she state that discourses should not be seen as wholly cohesive, since they always contain within them conflicting sets of statements (2003: 64). This also serves to illustrate the utility of maintaining a distinction between discourse and ideology when categorising stretches of text in this way. The transitivity analysis focuses on a single text, a decision motivated primarily by the desire to allow for a detailed, in-depth analysis that would not be possible if multiple texts were examined. Focusing in detail on a single text in this way facilitates the identification of intertextual links and ties within the text, whereby discourses are not only invoked, but reinforced or modified as the reading process progresses. The text selected was chosen from those texts in which the Talkative Woman discourse is globally topical, and from the Mail Online, since this is the most frequently indexed discursive grouping, and the Mail is the publication in which it most frequently occurs (see section 5 for full details). Other factors influencing the selection of this text are that it is not the most overtly sexist of the Mail articles in the dataset and purports to be a straight news text rather than an opinion piece. In this way, the selection of this particular text affords analysis of the subtler and hence more insidious discriminatory and exclusionary discourses that abound (Toolan, 2007: 94) within contemporary media discourse. Analysis is based on the model of transitivity developed by Halliday (1967,1968, 1985), but seeks to account not only for that which can be observed directly at the level of the clause, but also the myriad intertextual cues and traces by means of which both dominant discourses and the ideologies they articulate are made manifest. For this reason, the present study follows Toolan (1996) and Walsh (2001) in working with an extended definition of the text to include such intertextual and contextual factors that lie outside more traditional definitions of the term, but that may nevertheless play an important role in the construction of meanings by reader and analyst alike. 6. Analysis 6.1 Content Analysis As a result of the grouping together of relevant textual elements described above, a total of seven broad language and gender-related discursive groupings were identified in the dataset.

8 Newspaper Representations of the Female Speaking Subject 21 These groupings are not discrete categories, but rather display considerable degrees of overlap and interrelatedness. Nor are they unitary entities, but rather comprise elements of text that can be seen to draw on a variety of discourses or discursive strands situated in a range of sometimes complementary, sometimes contradictory ideological positions around a common discursive object. Moreover, these grouping are, as has already been discussed at some length, inherently interpretive categories the empirical validity of which should not be overstated. The discursive groupings identified and the number of texts in which each was observed at least once are illustrated figure 1, while an overview of each discourse with examples of the textual elements in which they were seen to be indexed is given in appendix 1. Figure 1: Number of texts in which discourses identified at least once As the chart illustrates, the Talkative Woman discourse is the most frequently represented language and gender discourse identified in the dataset. It would, perhaps, be more consistent to label this grouping as differences in amount of talk, since not all the textual elements that comprise this grouping, nor texts in which they were observed, are supportive of the notion that women talk more than men. However, texts and elements that contest the notion that women talk more than men are in a small minority in the dataset and, with the exception of one stretch of text taken from the website of the Telegraph, all appeared in texts published in The Guardian, a newspaper known for its relatively liberal editorial stance on many issues. Moreover, even these stretches of text, in contesting the discourse of the talkative women, also draw upon it, while the stereotype upon which it is based is one that has been widely observed in a range of discourses and text types (see below). For all of these reasons, it was decided to label this grouping the talkative woman discourse, despite the fact that this may suggest a more unitary ideological character than is actually the case. The long history of the stereotype of the talkative woman has been widely noted in the research literature (Swacker 1975, Talbot 2003, Cameron 2007), yet despite the overwhelming weight of evidence against it (James & Drakich 1993, Leaper and Ayers 2007), the stereotype of the talkative women has proved resilient. As Talbot notes, mere empirical evidence [ ] is unlikely to undermine the deeply held belief that women talk more than men, (2003: 480). A number of explanations have been offered for the persistence of this stereotype in the face of such overwhelming evidence against its veracity. Dale

9 22 Tony Fisher Spender famously suggested that women appear talkative because they are judged not in comparison with men but with silence (1980: 42), a view that echoes Ardener s assessment of women as a muted group a few years earlier (1975, 1978, cited in Cameron 2006: 13). However, such explanations seem to be based on hypotheses that it is simply not possible to test or verify in any meaningful way, and they arguably overlook the normative and potentially influential nature of the language and gender discourses and ideologies found in contemporary media texts. Before moving on to a more detailed analysis of the linguistic construction of the Talkative Woman discourse in one newspaper article from the dataset, there is one further quantitative observation that merits attention. The chart below shows the number of times any of the discursive categories are manifest at least once in the articles that appear in the online newspapers examined. This is set against the total number of texts each newspaper contributes to the dataset, and within which those elements are observed Express Guardian Independent Mail Mirror Observer People Sun Telegraph Times Instantiations of discursive categories No. of texts Figure 2: Occurrence of textual elements catalogued across newspaper titles Again, the need for caution in interpreting this information must be stressed, but the identification of 31 relevant textual elements in only 9 texts in the Mail Online must surely give some indication of the relative intensity with which gender and talk represents a preoccupation for this publication during the time period covered by the study. The Daily Mail was the first UK national newspaper to feature a women s section (MacDonald 1995: 76), and is currently the only national daily with more female than male readers - approximately 53% since 2007, according to the Mail s own figures (mailclassified.co.uk). This may go some way towards accounting for the extensive manifestation of gender-related discourses in the Mail articles contained in the data, but what it does not account for, and indeed renders all the more puzzling, is the nature of the ideological work potentially performed by such discourses in this publication, an account of which is given in the transitivity analysis that follows. 6.2 Transitivity Analysis In The Female Brain, the book that represents the focus of the article, 2 neuropsychiatrist Louann Brizendine claims that women produce on average 20,000 words of speech a day while men produce approximately 7,000, and it is this claim that represents the focus of the news text to be examined. In fact the claim was a spurious one, for which Brizendine provided no supporting evidence, and that runs counter to the findings of a significant body of research that, on average, there is little difference in the amount of talk produced by women and men (James & Drakich 1993, Leaper & Ayers 2007, Mehl et al. 2007). As Cameron notes, Brizendine s claim, which she describes as perfect material for soundbite science (2007: 19), subsequently caught the attention of copy editors around the world and

10 Newspaper Representations of the Female Speaking Subject 23 was as a result widely reported in the global press, a process that inevitably resulted in a degree of hardening as this unsupported claim quickly came to take on the appearance of scientific fact. After a sustained (and ongoing) internet based campaign by University of Pennsylvania professor Mark Liberman (Liberman, online resource), Brizendine removed the claim in its numerical form in subsequent print-runs, but maintains that women talk more than men in roughly the proportions suggested by the figures provided in the first edition (Brizendine 2006: 36). The article itself is accompanied by a photograph of a woman talking on the telephone (see MacRae 2006). Interestingly, the women s eyes are out of shot, so that the image represents not a photograph of an individual women, but rather a depersonalised representation of womanhood, in which the woman in view is little more than carrier of an attribute, in this case the ability or propensity to talk (Kress & van Leeuwen 1996: 50). The analysis below demonstrates how this representation finds a parallel in the transitivity choices evident in the text. Some 15 categories of participant were identified in this text, the two most frequently occurring of which are women and men. The verbal elements predicated of these two categories of participant were analysed for their status as process types (Halliday 2004: 172), and the results are given below verbal mental behavioral relational women men Figure 3: Process types attributed to women and men While only three verbal processes are attributed to men, eight are attributed to women. It could be argued that this merely reflects the focus of the article, Brizendine s claim that women speak three times as much as men. However, even when women and men are represented as being engaged in what amounts to dialogue, the way their participation is represented is markedly different: 5. It is something one half of the population has long suspected and 6. the other half always vocally denied. 7. Women really do talk more than men. That one half of the population refers to men and the other half to women, while not difficult to recover, is nowhere stated explicitly in the text. To arrive at a coherent interpretation the reader must make an inferential leap (Fairclough 1995: 122) in associating the verbal groups long suspected and vocally denied with men and women

11 24 Tony Fisher respectively, and in order to achieve this must recognise a cultural stereotype according to which men generally consider women to be excessively talkative, while the use of the verbs suspect and deny presupposes a shared assumption that talkativeness is a negativelyevaluated quality. Moreover, these verbs function interdiscursively, invoking here a discourse of criminality and detection or pursuit, which together with the adverbs long and vocally serve to cast women as elusive and tenacious transgressors, and men in a position of silent judgment. The verbs deny and suspect differ in another important way, in that the former is a verbal process and the latter a mental process, and this attribution of a mental rather than verbal process to men can be interpreted as a structured absence (Walsh 2001: 35), which serves to construct women as speaking subjects and men as non-speaking subjects: women talk while men think, women vocally transgress while men silently judge. So far, women have been constructed in this text as agentive in the verbal processes attributed to them. However, a closer reading reveals that this is not the only representational mode in operation, as can be seen in the following two stretches of discourse: 11. women also speak more quickly, 12. devote more brain power to chit-chat 13. and actually get a buzz out of hearing their own voices 21. the simple act of talking triggers a flood of brain chemicals 22. which give women a rush similar to 23. that felt by heroin addicts when 24. they get a high Clause 11 continues the pattern observed above, in which women are represented as agentive participants in a verbal process. However, this changes in clauses 12 and 13, the latter of which can be analysed as follows: 3 Subject Predicator Complement women actually get a buzz out of hearing their own voices Senser Process: mental Range: phenomenon The first verbal group, get a buzz out of can be paraphrased less idiomatically as enjoy or like, but the selection of this idiomatic construction is significant for at least two reasons. Firstly, this phrase contains within it the nominal element buzz, which in this context relates to the neurologically grounded experience of pleasure, while the verbal element get, which here is synonymous with receive, lacks volitionality to an extent that enjoy does not (Hopper & Thompson 1980), and is therefore more markedly non-agentive. In functional terms, women are thus cast in the passive role of Senser, the passive recipients of pleasurable experiences that are a function of neurological processes over which they have no control. Secondly, taken as a whole, this idiomatic construction functions interdiscursively (Fairclough 1992), drawing on a discourse of narcotic-like stimulation, an interpretation for which support can be found in corpus evidence. The British National Corpus contains only 10 instances of this phrase, but two of these relate to narcotic stimulation one with reference of caffeine, the other to magic mushrooms, while the Collins Wordbank returns 129 hits, of which 24 refer to narcotic stimulation, with reference to substances such as methadone, heroin, cocaine, marijuana and coffee. Indeed, this allusion is developed more fully in the second extract, 8 clauses later, when the cohesive link established by the reiteration of get a buzz (attributed

12 Newspaper Representations of the Female Speaking Subject 25 to women) in get a high (attributed to heroin addicts), the lexical frame reiteration (Halliday & Hasan 1976, Tannen 1989) and the lexical relation of synonymy that obtain all serve to establish an equivalence between women and heroin addicts. Thus, in this stretch of text, women are represented as the passive victims of neurologically determined, and socially deviant addiction. The denial of agency to women is maintained in the second extract where the NP women is indirect object complement in a relative clause that is embedded within the complement of the matrix clause. Subject Predicator Complement the simple act of talking Actor: force triggers Process: material Goal a flood of brain chemicals [which give women a rush similar to [that felt by heroin addicts when they get a high]] The third person plural verb give in the embedded relative clause selects the NP brain chemicals, rather than a larger stretch of discourse, as the anaphoric antecedent of the relative pronoun which, and thus as actor in this clause. Subject Predicator Indirect Object Complement Direct Object Complement which give women a rush similar to that felt by heroin addicts when they get a high Actor: force Process: material Beneficiary Goal The NP women is non-agentive beneficiary of the material process give so that, once again, women are represented as the passive recipients of the intense heroin-like euphoria they allegedly experience as a result of talking. However, closer examination reveals other processes at work in the background, and this in turn suggests that the denial of agency to women is both incomplete and selective. At this point it is necessary to return to clause 13 in the first extract. It was suggested that the verbal group get a buzz out of is a mental process that renders the grammatical subject women as Senser, and therefore as non-agentive in the process predicated upon them. However, get a buzz is experientially dependent upon the process of hearing, the inherently process like nature of which is here somewhat obscured, or at least significantly backgrounded, by the nominalization hearing their own voices in which it is embedded: hearing Process: mental their own voices Phenomenon: range

13 26 Tony Fisher This implicit process in turn is dependent upon and presupposes the act of speaking, a verbal process in which women perform the function of Sayer, a far more agentive role than that of passive Sensor. A similar situation can be observed in the second clause complex, where women are the passive recipients of a rush [ ] that is produced by brain chemicals. The actor, or force, in this clause, the simple act of talking, is a nominalization that similarly obscures, or again at least backgrounds, the verbal process of talk, in which the elided Sayer, women, is again agentive. Although in both cases severely backgrounded, the agency associated with women is further reinforced by, and in turn reinforces, the underlying discourses of elicit pleasure and narcotic stimulation on which the text draws; like heroine addicts, women are passive participants in the neurological processes underlying their alleged addiction, but agentive in the verbal behaviours that lead to those processes. This parallel attribution of participant roles to women and heroine addicts is an example of negative comparison, a discourse feature van Dijk argues can be implicated in the ideologically based description of Others [ ] by comparing the target person or outgroup with a generally recognised Bad person or outgroup (1995: 154-5). This echoes the construction of women as other witnessed in clause 6 (Kress & van Leeuwen 1996: 138; O Keefe 2002: 101), where women are described as the other half of the population, whist also reinforcing the construction of women as transgressors identified in the first extract examined above. The analysis above thus suggests a double articulation of participant roles that results in a dual representation of women at two different levels of signification, one explicit and one implicit. While the surface representation of women as transgressors is to some extent ameliorated by their representation also as victims, this latter is arguably undermined by their construction in the role of agentive transgressors implied in processes that are backgrounded and obscured by the transitivity choices evident in the text. The net result is that women are represented not so much as non-agentive, but as lacking in control, unable to curb their appetites for talk, and as socially deviant in much the same way as heroin addicts looking for the next fix. 7. Discussion: the Panoptical Press Writing in the late 1980s, Sandra Lee Bartky notes that women s behaviour is less rigidly regulated now than in the past, and argues that these changes have resulted in a modernisation of power over women parallel to, though much later in coming than, the dissipation of power away from the figure of the sovereign throughout Europe that Foucault describes in The History of Sexuality, Vol. 1 ([1978] 1990). Where patriarchal power was once exercised directly by such figures as the father, the husband and the parish priest, for many women, such top-down effects of power are no longer such a salient part of everyday experience (1988: 80). Following Foucault (1972), Bartky draws on the panopticon metaphor to account for the new regime of surveillance and power in which women are the principles of their own subjection: The women who checks her makeup half a dozen times a day to see if her foundation has caked or her mascara has run, who worries that the wind or rain may spoil her hairdo, who looks frequently to see if her stockings have bagged at the ankle or who, feeling fat, monitors everything she eats, has become, just as surely as the inmate of the panopticon, a self-policing subject, a self committed to a relentless self-surveillance. (1988: 81) But self-surveillance alone is not enough; essential to the successful functioning of the panopticon is the ever-present possibility of external surveillance, and this implies the existence of some form of authority figure, a top sergeant in the disciplinary regime of

14 Newspaper Representations of the Female Speaking Subject 27 femininity (74). Bartky considers a number of candidates, including parents, teachers, a generalised panoptical male gaze and the media, but concludes that ultimately the disciplinary power that inscribes femininity in the female body is everywhere and nowhere; the disciplinarian is everyone and yet no one in particular. (ibid) Bartky is trying here to capture the undefined, omnipresent and multivalent nature of power and the disciplinary procedures that effect women in particular, but it could be argued that in dispersing disciplinary power everywhere, she is vulnerable to the same criticisms made of Foucault s modern power (Hall 1988: 52, Newman 2005: 53) in that she overlooks the fact that some sources of power are more influential than others. The seven language and gender discourses identified in the data examined above articulate a range of injunctions and prescriptions directed at a largely female readership. In these are articulated a variety of language and gender ideologies that variously represent women as preoccupied with inconsequential conversational topics, unreasonable and inconsistent in their attitudes to compliments from men, obsessed with emotions, and above all too talkative. Sometimes women are represented in a positive light: they are more polite, possess superior linguistic and communicative skills, and are possessed of an interactional style better suited to conflict resolution than that of any man. However, while such qualities are positively evaluated in the private sphere, they are not so well received in the public sphere, where, for example, women s fluffy styles are considered inappropriate for business communication (Times Online, January 31, 2007). And if women should abandon their fluffy styles in favour of stereotypically male discursive strategies, they become shedevils in the boardroom (Mail Online, June 12, 2008), quickly finding themselves in the double bind noted by Lakoff (1975) and others. It seems, then, that whatever discursive or linguistic strategies and resources they employ, women either transgress or risk transgressing. However, as Cameron (2003: 449) suggests in her account of the double discourse that generally characterises representations of language, debates about language are rarely (if ever) only about language, but rather pertain to wider social concerns. In this sense, it is arguably the case that woman s real crime is not simply focusing on inconsequential conversational topics, nor even talking too much, but a persistent disobedience as subject in a patriarchal system, a subject lacking in selfcontrol, and who is ineffective in her own self-surveillance. On this reading, language and gender discourses such as those identified above can be seen to function not only in the articulation of normative language and gender ideologies, but as symbolic of a wider, postfeminist inflected concern with shifting gender identities and relations (Gill 2007: 255, see also Hall & Rodriguez 2003), and an insidious reminder of the ever constant threat of the patriarchal panoptical gaze. 8. Conclusion The content analysis presented above illustrates the variety of language and gender discourses that can be observed in the dataset, while the qualitative analysis reveals some of the ways in which one language and gender discourse, that of the Talkative Woman, is made manifest in the transitivity choices evident in the text. This analysis also demonstrates the inherently interdiscursive nature of text and discourse evident in the construction of women, in one particular text, not only as subjects possessed of an insatiable appetite for talk, but as deviants or transgressors who display a pathological addiction to talk comparable to that experienced by some drug addicts to heroin. Furthermore, it was revealed how the transitivity choices made served sometimes to background particular processes, with the result that even where women were represented as non-agentive in the pleasure they allegedly experience as a result of talk, they remained agentive in the act of talking that fuelled the pathological